[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [soa-rm] My stab at an SOA-RM concept map
So, here is another go... For any service there is an expectation as to how to use the service, and what the results of using the service might be. Loosely, this is what most people (I humbly suggest) mean by the semantics of the service. The data model and the process model would normally be thought of as part of the 'how to use the service' aspect of its semantics. The concept of policy is normally defined to be a constraint on the behavior of entities. Taken at its most liberal, this includes the above: the service is constrained to behave the way it is supposed to. However, in our discussions, I think that we went a stage further in nailing down what policy meant: that it was a constraint on the behavior of a service (or state, or the service consumer etc. etc.) that is unilaterally declared by one or more of the participants of a service usage scenario. Finally, a contract (or agreement if contract is too loaded), represents a policy that has been agreed to by at least two participants; but more generally, by all affected participants. I think that it is interesting that in this analysis, it seems impossible to avoid bringing in people (and therefore service consumers etc.) into the picture of service oriented systems. However, you cannot fully account for this picture unless you also bring in the concept of a real world effect. I.e., business semantics. The reason is that a policy without a policy owner (sometimes called the policy subject I believe) is an oxymoron: it is the owner of the policy that distinguishes a policy from just any predicate. But owning is a real-world concept (as opposed to the Java concept of inheritance which is not of this world :) Frank On Jun 1, 2005, at 9:56 AM, Peter F Brown wrote: > Another go: > Looking back at another earlier posting from Ken, maybe we can say > that "the > semantics of a service *are exposed by* the policy".... > > -Peter > > -----Original Message----- > From: Peter F Brown [mailto:peter@justbrown.net] > Sent: 01 June 2005 18:52 > To: 'SOA-RM' > Subject: RE: [soa-rm] My stab at an SOA-RM concept map > > <quote> > Not sure about the direction of some of these arrows. > </quote> > > It depends on what visualisation tool/method you are using...back > to the > confcall debate... ;-) > > I do not understand the arrows to mean anything more than > illustrate that > the words imply a sentence structure of SVO (Substantive-Verb-Object) > > More precisely it will be an ontology that constrains meaning, not > semantics, but in our context I'm wondering whether we should make > *any* > association between semantics and policy, except to say textually > (and this > is not a formal proposal for wording) "Policy needs to be > understood as a > mix of explicitly defined service description and the meanings > implicit in > other objects and associations within the model..." I know, it's a bit > flaky... > > -Peter > > -----Original Message----- > From: Francis McCabe [mailto:fgm@fla.fujitsu.com] > Sent: 01 June 2005 18:31 > To: Behera, Prasanta > Cc: Matthew MacKenzie; peter@justbrown.net; SOA-RM > Subject: Re: [soa-rm] My stab at an SOA-RM concept map > > Not sure about the direction of some of these arrows. > > Assuming that > > A ---part of--> B > > means A is a part of B, then I think > > Data Model ---part of--> Semantics > > reflect my pov. > > Continuing, this.... > > Policy ---part of--> semantics > > Semantics ---defined-by--> Ax. x ---part of--->Semantics > > Actually, I am not that part-of is the correct predicate. View-of > seems > closer to the truth. > > For example, what is the data model? The question is non-trivial > because > there is a close synergy between the data model and the process > model. And a > part-of relationship suggests separable components. > > Frank > > > > > On Jun 1, 2005, at 9:09 AM, Behera, Prasanta wrote: > > >> How about this? >> >> Semantic ---part of ---> Data Model >> Semantic ---constrain --> Policy >> Semantic ---define------> Policy >> Semantic ---constrain---> Service (not too sure for this one). >> >> Does this work? >> >> Thanks, >> /Prasanta >> >> -----Original Message----- >> From: Francis McCabe [mailto:fgm@fla.fujitsu.com] >> Sent: Wednesday, June 01, 2005 8:55 AM >> To: Matthew MacKenzie >> Cc: Behera, Prasanta; peter@justbrown.net; SOA-RM >> Subject: Re: [soa-rm] My stab at an SOA-RM concept map >> >> Actually, I am not happy with the constrain word here. >> >> The data-model, for example, is not constrained by the semantics of >> the service -- it *is* part of the semantics. >> >> Frank >> >> On Jun 1, 2005, at 8:50 AM, Matthew MacKenzie wrote: >> >> >> >>> Ok, I'm happy with constrain as well. >>> >>> -matt >>> Behera, Prasanta wrote: >>> >>> >>> >>> >>>> That would be the term of choice for me too. >>>> (Sorry. Didn't get a chance respond early) >>>> >>>> Thanks, >>>> /Prasanta >>>> >>>> >>>> -----Original Message----- >>>> From: Peter F Brown [mailto:peter@justbrown.net] Sent: Tuesday, May >>>> 31, 2005 6:56 PM >>>> To: 'SOA-RM' >>>> Subject: RE: [soa-rm] My stab at an SOA-RM concept map >>>> >>>> Surely semantics "constrain" rather than "govern" or "define"? >>>> Specifically, >>>> they constrain meaning within a specified domain ontology >>>> >>>> -Peter >>>> >>>> -----Original Message----- >>>> From: Behera, Prasanta [mailto:pbehera@visa.com] Sent: 31 May 2005 >>>> 20:03 >>>> To: Matthew MacKenzie; Francis McCabe >>>> Cc: SOA-RM >>>> Subject: RE: [soa-rm] My stab at an SOA-RM concept map >>>> >>>> #1: Can u expand a little bit on the semantics of "govern"? >>>> Semantics >>>> "define" and "govern" policy. >>>> >>>> #2: The relationship between "Data Model" and "Services" -- should >>>> it be more of a association type (line instead of a arrow) >>>> >>>> Thanks, >>>> /Prasanta >>>> >>>> -----Original Message----- >>>> From: Matthew MacKenzie [mailto:mattm@adobe.com] >>>> Sent: Tuesday, May 31, 2005 9:13 AM >>>> To: Francis McCabe >>>> Cc: SOA-RM >>>> Subject: Re: [soa-rm] My stab at an SOA-RM concept map >>>> >>>> Having some problems with arrows in this tool, but I added another >>>> relationship to contract (attached). >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>> >>> >>> >>> >> >> >> > > > >
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]